If your using EFS, enter the highest return you've had...

[BakedWafer]

Well, let's just hope I don't loose on trade one then... I'll respond more later.

J

[TraderJO]

And then....

we have this. An example of excellent back test results, but I'm having a devil of a time getting the same thing in Real Time. Close, but no cigar, as they say. This is what happens if your logic enters trades at the open of the signal bar, instead of the close. Actually, real time testing is close to this, but you get some false entries when a bar is a sell at it's open, and a buy at it's close.

[spike500]

Re: And then....

Originally posted by TraderJO
we have this. An example of excellent back test results, but I'm having a devil of a time getting the same thing in Real Time. Close, but no cigar, as they say. This is what happens if your logic enters trades at the open of the signal bar, instead of the close. Actually, real time testing is close to this, but you get some false entries when a bar is a sell at it's open, and a buy at it's close.

You can only trade at the close of the bar, because only at the close you are sure that the graphs are as they are in backtesting.
Indeed a long signal can still change 20 times before the close of the bar so the only correct way to trade is on the close of the bar.

I find it strange that so many people can produce terrible equity curves in backtesting, but in reality it seems to be "a little bit" different. Emotions canb have some influence, but i think that a bad methode of testing a system is the main reason why results are different between testing and real trading.

[spike500]

TraderJO

zip worked well.

few questions:
margin per contract only 40$ ? is this emini S&P? My margin is intraday 400$ per mini

commission pe contract only 0.16$? my commission is at least 2.5$

Your results look quite reasonable.
Only the drawdown is too big to me. That can be controlled by a lower leverage.

[TraderJO]

Answers...

Spike,

As I've discussed in previous posts, all the back test parameters must be multiplied or divided by the point value of the instrument you're trading when trading futures. In the case of the ES it is $50 per point. So my commission per round turn when using my trading platform in auto execute mode is up to $8.00, so 8/50 = .16 Same with the commission 50x40 = 2,000. For trade size I'm also multiplying by 50 so 3 contracts = 150 entered.

This particular system enters with three contracts, has two profit targets, and lets the third one run with a trailing stop. It rarely misses a nice runner, but does get whacked a bit in choppy markets.

I can't understand how the back tester calculates maximum drawdown, as the report shows a huge maximum drawdown and account size required. But, when you look at the equity curve it appears to have a 4K max dip in the curve. Any ideas on this? You wrote about drawdown calculations earlier, but it didn't make sense to me. I appreciate your input.

[spike500]

TraderJO

I agree that your summary looks suspicious to me. Indeed the drawdown says minus 198465$ but i don't see this drawdown anywhere in your equitycurve. According to your equity curve your drawdown is very small.
The question now is: what's right and what's wrong?
The equitycurve or the summary.

I'm still waiting after the first performance summary that is correct after checking ist manually. I have a Tradestation and there the same problem exist. I only thrust those things that i calculated myself and manually.

About my explanation on drawdown. What would you like me to explain more profoundly?

[BakedWafer]

I have taken almost everything in to account. I think :/ I even overcompensated on some stuff to make it more realistic.

[TraderJO]

Drawdown calculations

Indeed....what's right and what's wrong. Well, this wouldn't be the first glitch I've found in various Advanced Charting features. If I go over the trade by trade results, it seems that the equity curve is correct, but I could have missed something. At this point I'm going to assume the problem lies in the report generator.

After reviewing your post on drawdowns, I guess I do understand it completely, and agree whole heartedly. When one starts trading a system with real money, the best assumption to make is that you will begin at a peak on the curve, and experience a loosing streak straight away. This has happened to me in the past with real money, and it's hard to accept and keep on trading the system. So....test, test, test, and then grit your teeth and go for it !

[derekg]

Re: And then....

Originally posted by TraderJO
This is what happens if your logic enters trades at the open of the signal bar, instead of the close. Actually, real time testing is close to this, but you get some false entries when a bar is a sell at it's open, and a buy at it's close.

Agreed. In fact when backtesting I don't even use the close of the same bar, preferring the open of the next bar instead. Seems more realistic.

[Doji3333]

I agree with Derek..

I check NEW BARS (when they start forming on the chart) and compare my entry conditions for the -1,-2,-3 bars (as many as I need), then issue an order to BUY/SELL with "Strategy.MARKET, Strategy.THISBAR" conditions.

This way, when I convert the code from BT to RT mode, my logic does not have to change (much) at all. Everything is there to address the issues of conversion.

But generally, you have to be cautious of using "forward looking" conditions in your code. Could cause lots of problems.

B

[derekg]

That's a good idea. I was noticing problems with Strategy.MARKET, Strategy.NEXTBAR conditions. Mostly visual glitches like signals appearing and disappearing. More of an annoyance than anything but I would like to rewrite the code when I get a chance.

[gspiker]

Indeed a long signal can still change 20 times before the close of the bar so the only correct way to trade is on the close of the bar.

Depends on the system and what it is trying to achieve. Some systems don't require a bar close, and will produce worse results if you wait for such to happen. There are few absolute rules such as this in trading.

The backtester can even take into account prices before the close of a bar, if you code it correctly.

Garth

[Doji3333]

Update

Hey Folks...

So far, this week has gone good (as far as development of code) and kinda sideways for my system.

The RT code is doing well. It's been running now for 3 days and the results are nearly identical to the BT code. The only difference is when limit orders actually get filled. So, I'm addressing this with minor modifications to the system.

The BT code is still doing fine, but I have some ideas that may help it out.

This week, though, has been pretty much a "WASH" as far as the results - a loss of $245.20..

Monday : -$403.60 (closed trading after 4 losses)
Tuesday : +$1277.20
Wednesday: +$15.59
Thursday : -$1134.40 (closed trading after 4 losses)

So, let's see what happens tomorrow and see if this system can show a profitable week.

IMHO, a flat week like this does not really matter over the long term of the system - you're going to have some good weeks and some bad weeks - part of the game.

B

[derekg]

Originally posted by BakedWafer
I have taken almost everything in to account. I think :/ I even overcompensated on some stuff to make it more realistic.

Nice looking results. Seems like you get stopped out a lot, but when you're on the right side it really lets the profits run. I wonder why it was having trouble in April. Market conditions seemed a lot like February but your strategy didn't have any problems that month.

[spike500]

Originally posted by gspiker
Depends on the system and what it is trying to achieve. Some systems don't require a bar close, and will produce worse results if you wait for such to happen. There are few absolute rules such as this in trading.

The backtester can even take into account prices before the close of a bar, if you code it correctly.

Garth

Garth,

i don't agree with you. You don't have to wait for the bar to close only if you trade on a pricelevel that has to be reached. If the price is met you can trade because you are sure that the condition to trade is met.

But if you trade on indicators you always have to wait until the bar has closed. Otherwise you are never sure that the condition to trade will be met.
If you backtest a system and rely on the results you have to remember that the backtest only knows the value of an indicator on the close of that bar. A backtest will never take all the different values inside of that bar. If you trade in real time you have to follow the same rules as your backtest. Otherwise backtesting is useless. The results of the backtest are realized in following certain rules, so you have to follow these rules. If you don't do this it is like testing something and doing in realtime something else. The results will never be the same because you did two different things. So something that was profitable in backtesting could than become a nightmare.